Re: [Wrapper-user] (no subject)

[Leif M. <le...@ta...>]

Walling,
The JVM appearing to be hung state could be caused by a number of things.
Including actual JVM hangs, memory swapping problems or high loads. There
have been a number of related bugs fixed in later versions. 3.0.3. The new
timer mechanism as of 3.1.2 now handles heavily loaded systems much more
reliably.

That said however, the fact that classes are not being found after a JVM
restart is odd. What state is the system in at that point. do you have to
restart to system to get things working again? This is not a problem that I
have ever heard of before. I wonder if the JVM is being frozen by something
related to the jar / class files not being found after a restart.

If you can catch this with wrapper.debug=true set, I might have a little
more information for you as to the cause.

I just noticed that the jar files are being loaded from a network drive.
Is it possible that the network connection is dropping? That could possibly
be causing the JVM to hang when it tries to access a class or other
resource it had not yet loaded. That would also explain the failure to
reload the classes on a JVM reload.
I have seen cases where a JVM would crash if I overwrote jar files
from Ant or something while the JVM was running. This could be a related
problem.

Cheers,
Leif

dp...@ro... wrote:
> Subject: JVM appears hung, and SOMETIMES java class not found
>
>
>
> I have had the following problem 4 times since the start of 2005, and I'm
> hoping someone can suggest what I should be looking at.
>
> (This situation is in our production environment which has very tight
> controls over changes, so I am certain that nothing has changed in the
> source, or security between when this works and when it fails)
>
> Frequently, (about once a week) we get the messages
> ERROR  | wrapper | date time | JVM appears hung: Timed out waiting for
> signal from JVM. ERROR  | wrapper | date time | Java Virtual Machine did
> not exit on request, terminated INFO  |jvm 4  |  date  time | Launching a
> JVM
>
> and usually the next 2 messages are:
>
> INFO  | jvm   | date time | Wrapper (Version 3.0.3)
> INFO  | jvm   | date time |
>
>
> 4 times this year however, we get the following:
>
> ERROR  | wrapper | date time | JVM appears hung: Timed out waiting for
> signal from JVM. ERROR  | wrapper | date time | Java Virtual Machine did
> not exit on request, terminated INFO  |jvm 4  |  date  time | Launching a
> JVM
> ERROR  | wrapper | date time | The java class is not found:
> com/solcorp/pathfinderwrapper/WrapperStartStopApp
>
> The "Launching a JVM" and "class is not found" messages repeats 5 times,
> and then another message, (which I don't have handy right now) indiates
> that there have been 5 failed starts in a short period of time, and the
> service is stopping.
>
> >From here, we have to stop and restart our application.
>
>
> Why does it successfully restart most of the time, but not all of the
> time?
>
> Our envionment:
>
> We have 2 Win2003 servers in a websphere (5.1) cluster.
> Both servers have 3 NT services running with the wrapper.
> The config files for the wrappers are sitting on a NAS (network attached
> storage) drive, and the same files are shared between the 2 servers. (if
> your familiar, this is part of an Ingenium instalation, and my expertise
> is from the Mainframe world).
>
>
> Thanks in advance  for any suggestions 
>
> King Walling
>
>
>
>
>